[master] 668e051db Typo

[Dridi Boukelmoune]

> -Using ``VRT_priv_task()`` to maintin per object instance state is a
> +Using ``VRT_priv_task()`` to maintain per object instance state is a
>  convenient yet unofficial interface which was not originally intended
>  for this purpose and will likely be replaced with a more suitable
>  interface.

So... This has been around since 2014, almost 4 years. I'm not
comfortable telling this is not blessed by $ABI vrt.

Either it is, or we can introduce a new "VRTC" prefix for VRT Candidate
functions, which is fine since this release already has a major VRT bump.

And 6.1 could introduce the concept of candidate APIs, especially since
slink already documented the $ABI rule of thumb, we now have a place
to explain the difference between VRTC and VRT if we go that route (and
we can add it to the release notes for 6.1 too).

I like the idea of giving ourselves an escape hatch when we aren't
sure what a VRT interface will look like, or when we want to expose it
while parts are still expected to move before it stabilizes. Less
major bumps ahead possibly.

Thoughts?

Dridi